Meet LG AI Research at NVIDIA GTC 2025!

Meet LG AI Research at NVIDIA GTC 2025! | by LG AI Research | Medium

Sitemap

Open in app

Sign up

Sign in

Medium Logo

Get app

Write

Search

Sign up

Sign in

Meet LG AI Research at NVIDIA GTC 2025!

LG AI Research

4 min read

Mar 28, 2025

--

Listen

Share

The LG AI Research booth at GTC 2025

The world’s largest AI developer conference, NVIDIA GTC 2025, is taking place in San Jose, California, from March 17 to 21 (local time). The event has brought together numerous global IT companies and AI experts, and LG is also present at the scene.

LG has set up a booth to showcase its latest AI advancements, featuring four key solutions: EXAONE Deep, ChatEXAONE, EXAONEPath, and DDU(Deep Document Understanding) technology. Attendees have the opportunity to experience these innovations firsthand through live demonstrations, drawing large crowds to the booth.

Curious about the latest AI innovations from LG AI Research?Here’s a look at the key technologies and live demos available at the LG booth!

EXAONE Deep

Press enter or click to view image in full size

Visitors experiencing a live demo at the LG AI Research booth at GTC 2025

At the booth, LG AI Research introduced EXAONE Deep, a powerful Reasoning AI capable of competing with global reasoning-specialized models. EXAONE Deep is a high-performance reasoning model that understands mathematical logic, infers scientific concepts, and solves programming problems. Immediately after its release in March 2025, the 32B model was listed in the “Notable AI Models” by EPOCH AI, a U.S. nonprofit research institution, proving its applicability in specialized knowledge fields. From EXAONE Deep to EXAONE 3.5, LG AI Research has been the only domestic company to have its models listed over the past two years, marking a significant achievement.

→ Try EXAONE Deep

Press enter or click to view image in full size

A spectator reviewing the performance result data of EXAONE Deep

Also, at the booth, LG showcased EXAONE 3.5-powered solutions for automated code generation and data creation. The AI-powered code generation tool quickly produces code from natural language requests, allowing users to focus on more complex tasks. Meanwhile, the automated data generation solution ensures high-quality training data, with built-in monitoring to filter out low-quality outputs.

ChatEXAONE

LG also presents ChatEXAONE, an AI assistant designed to enhance enterprise productivity. ChatEXAONE understands business documents, summarizes key insights, analyzes data, and even retrieves real-time online information for the most relevant responses.

Press enter or click to view image in full size

ChatEXAONE’s Deep & Dive feature interface

A new addition at GTC 2025 is the ‘Deep & Dive’ feature. The Deep function provides a transparent breakdown of AI’s reasoning steps in generating answers, while the Dive function allows users to customize their information sources, selecting from international references, videos, research papers, and more. This level of customization ensures a highly personalized ChatEXAONE experience.

EXAONEPath

LG is also developing AI models tailored for specific industries. One such model, EXAONEPath, is optimized for processing histopathology images. These images, often as large as 20k x 20k pixels, feature unique color variations and require specialized processing due to the limited types of objects involved (e.g., nuclei, cytoplasm).

EXAONEPath has garnered significant attention for achieving performance on par with global competitors while reducing model size to just 1/10 of traditional solutions. This efficiency has impressed AI developers and industry professionals alike.

The model is expected to revolutionize the bio sector, drastically reducing the processing time of genetic testing from up to two weeks to just a few days, ultimately saving both time and costs. EXAONEPath is now open-source, allowing researchers to utilize and further develop the technology.

→ EXAONE Path Paper

DDU (Deep Document Understanding)

LG is also showcasing DDU technology, designed to help AI comprehend various types of information embedded within documents, such as tables, charts, and molecular structures.

By applying DDU to different industries, AI can extract key insights from graphs and generate valuable analytical results. At GTC 2025, LG is hosting a live demo of its DDU technology, offering visitors a glimpse into a future where AI transforms document-based insights.

Beyond GTC 2025, LG is committed to engaging with users and global AI researchers to showcase its AI technologies and exchange insights. Stay tuned for more innovations as LG continues to shape the future of AI!

Gtc

LG

Lg Ai

Exaone

Nvidia

Written by LG AI Research

6 followers

1 following

From powerful foundation models to advanced agent systems, we are building what’s next in AI — redefining the future through innovation and real-world impact.

No responses yet

Help

Status

About

Careers

Press

Blog

Privacy

Rules

Terms

Text to speech